[ofa-general] [RFC][PATCH] IPoIB-cm: release connection if post_send drain WR failure

Roland Dreier rdreier at cisco.com
Wed Jun 25 11:51:36 PDT 2008


 > I saw it happened from below ipoib debug output on testing ehca2:
 > 
 > static void ipoib_cm_start_rx_drain(struct ipoib_dev_priv *priv)
 > {
 > ....
 >         /*
 >          * QPs on flush list are error state.  This way, a "flush
 >          * error" WC will be immediately generated for each WR we post.
 >          */
 >         p = list_entry(priv->cm.rx_flush_list.next, typeof(*p), list);
 >         if (ib_post_send(p->qp, &ipoib_cm_rx_drain_wr, &bad_wr))
 >                 ipoib_warn(priv, "failed to post drain wr\n");

Was this with a kernel that had 863fb09f ("IB/ehca: Prevent posting of
SQ WQEs if QP not in RTS") but not 088af154 ("IB/ehca: Reject send WRs
only for RESET, INIT and RTR state")?  For example something between
2.6.25 and 2.6.26-rc4?

 - R.




More information about the general mailing list